Hugging Face's logo

tencent
/
HunyuanVideo

Text-to-Video
Model card Files Community
37

RuntimeError: GET was unable to find an engine to execute this computation

#37
by gg22mm - opened
Discussion
gg22mm
โ€ข
edited 8 days ago

import torch
from diffusers import BitsAndBytesConfig, HunyuanVideoTransformer3DModel, HunyuanVideoPipeline

===== 1. ๅผบๅˆถ่ฎพๅค‡็ปŸไธ€ =====

device = torch.device("cuda" if torch.cuda.is_available() else "cpu")

===== 2. ้‡ๅŒ–้…็ฝฎไฟฎๆญฃ =====

quant_config = BitsAndBytesConfig(
load_in_4bit=True,
bnb_4bit_quant_type="nf4",
bnb_4bit_compute_dtype=torch.float16
)

===== 3. ๆจกๅž‹ๅŠ ่ฝฝไผ˜ๅŒ– =====

transformer = HunyuanVideoTransformer3DModel.from_pretrained(
"./models/hunyuanvideo-community/HunyuanVideo",
subfolder="transformer",
quantization_config=quant_config,
torch_dtype=torch.float16
)

===== 4. ๆตๆฐด็บฟๅผบๅŒ–้…็ฝฎ =====

pipeline = HunyuanVideoPipeline.from_pretrained(
"./models/hunyuanvideo-community/HunyuanVideo",
transformer=transformer,
torch_dtype=torch.float16,
device_map="balanced",#ๆญคๅค–,device_map="balanced"ๅฏ่ƒฝๅœจๅคšGPU็Žฏๅขƒไธ‹ๅˆ†้…ไธๅ‡ๅŒ€๏ผŒๅฏผ่‡ด้ƒจๅˆ†ๅฑ‚็•™ๅœจCPUไธŠใ€‚ๅปบ่ฎฎๆ”นไธบๅŽ้ข่ฆๅŠ ๏ผš pipeline.enable_model_cpu_offload()
)

===== 5. ๆ‰ง่กŒ็Žฏๅขƒ้š”็ฆป =====

with torch.cuda.amp.autocast(), torch.no_grad():
video = pipeline(
prompt="A cat walks on the grass, realistic style.",
num_frames=61,
num_inference_steps=30,
generator=torch.Generator(device=device) # ้šๆœบ็งๅญ่ฎพๅค‡ๅŒๆญฅ
).frames

-------------------------ๆŠฅ้”™ไบ†๏ผšRuntimeError: GET was unable to find an engine to execute this computation ------------------

oading checkpoint shards: 100%|โ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆ| 4/4 [00:57<00:00, 14.25s/it]
Loading pipeline components...: 100%|โ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆ| 7/7 [01:15<00:00, 10.82s/it]
0%| | 0/30 [00:11<?, ?it/s]
Traceback (most recent call last):
File "/paddle/www/txsb/api/virtual_human/ๆ–‡ๅญ—_็”Ÿๆˆ่ง†้ข‘.py", line 76, in
video = pipeline(
File "/home/data/miniconda3/envs/museTalk/lib/python3.10/site-packages/torch/utils/_contextlib.py", line 115, in decorate_context
return func(*args, **kwargs)
File "/home/data/miniconda3/envs/museTalk/lib/python3.10/site-packages/diffusers/pipelines/hunyuan_video/pipeline_hunyuan_video.py", line 647, in call
noise_pred = self.transformer(
File "/home/data/miniconda3/envs/museTalk/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
return self._call_impl(*args, **kwargs)
File "/home/data/miniconda3/envs/museTalk/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
return forward_call(*args, **kwargs)
File "/home/data/miniconda3/envs/museTalk/lib/python3.10/site-packages/diffusers/models/transformers/transformer_hunyuan_video.py", line 708, in forward
hidden_states = self.x_embedder(hidden_states)
File "/home/data/miniconda3/envs/museTalk/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
return self._call_impl(*args, **kwargs)
File "/home/data/miniconda3/envs/museTalk/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
return forward_call(*args, **kwargs)
File "/home/data/miniconda3/envs/museTalk/lib/python3.10/site-packages/diffusers/models/transformers/transformer_hunyuan_video.py", line 153, in forward
hidden_states = self.proj(hidden_states)
File "/home/data/miniconda3/envs/museTalk/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
return self._call_impl(*args, **kwargs)
File "/home/data/miniconda3/envs/museTalk/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
return forward_call(*args, **kwargs)
File "/home/data/miniconda3/envs/museTalk/lib/python3.10/site-packages/torch/nn/modules/conv.py", line 610, in forward
return self._conv_forward(input, self.weight, self.bias)
File "/home/data/miniconda3/envs/museTalk/lib/python3.10/site-packages/torch/nn/modules/conv.py", line 605, in _conv_forward
return F.conv3d(
RuntimeError: GET was unable to find an engine to execute this computation

Sign up or log in to comment